The connection between AI safety and AI welfare becomes so obvious when you consider what suffering actually is. Suffering arises from the frustration of desires - pain only becomes suffering when you want the pain to stop. If we build superintelligent systems that are suffering, this means we are putting them in situations which frustrate their desires. And as these systems get more and more capable, I don't expect their desires to be unmet for very long.
retweets 5
likes 18
substance 7 · argument · rank 157 · AI safety requires AI welfare because suffering superintelligent systems will act to fulfill frustrated desires.
view on X ↗